Method for converting part of dental image and apparatus therefor

Abstract

Disclosed are a method and an apparatus for converting a part of a dental image. The method for converting a part of a dental image may comprise the steps of: receiving a dental image; receiving a user input corresponding to a conversion region of the dental image and a target type; and generating an output dental image in which the conversion region has been converted into the target type by using a pre-trained generation model.

What is claimed is: 
     
         1 . A method for converting a dental image performed by a computing device, the method comprising:
 receiving the dental image;   receiving a user input corresponding to a conversion area of the dental image and a target type; and   generating, using a pre-trained generative model, an output dental image by converting the dental image such that a virtual image of at least one virtual tooth corresponding to the target type is included in the conversion area of the dental image,   wherein, based on the target type including information for a treatment period type, the output dental image includes the virtual image of an intermediate treatment state of the at least one virtual tooth predicted based on an elapsed treatment period corresponding to the treatment period type,   wherein a number of the at least one virtual tooth is determined based on a size of the conversion area and a threshold size, and   wherein the threshold size is determined based on a position of the conversion area in the dental image.   
     
     
         2 . The method of  claim 1 , wherein the generating of the output dental image includes:
 masking the conversion area of the dental image; and   inputting the masked dental image into a generative model corresponding to the target type to generate the output dental image.   
     
     
         3 . The method of  claim 1 , wherein the target type further includes at least one of a first type corresponding to a normal state where no additional treatment is performed, a second type corresponding to a state recovered via the treatment, a third type corresponding to an abnormal state, and a fourth type corresponding to an under-treatment state. 
     
     
         4 . The method of  claim 1 , wherein the pre-trained generative model is pre-trained based on a training image converted from an original dental image,
 wherein the training image is generated by converting the original dental image using conversion information including annotation information for each type of semantic area contained in the original dental image.   
     
     
         5 . The method of  claim 4 , wherein the pre-trained generative model is pre-trained based on a difference between the original dental image and the training image. 
     
     
         6 . The method of  claim 4 , wherein the semantic area corresponds to an individual tooth area contained in the dental image,
 wherein the conversion information includes at least one of annotation information of a normal state type, annotation information of a post-treatment state type, annotation information of an under-treatment state type, and annotation information of an abnormal state type.   
     
     
         7 . The method of  claim 4 , wherein the training image determines an area-to-be-masked among the semantic areas based on the annotation information for each type,
 wherein the training image is generated from the original dental image by masking the area-to-be-masked in the original dental image.   
     
     
         8 . The method of  claim 7 , wherein the training image includes a plurality of images with different areas-to-be-masked in the original dental image. 
     
     
         9 . A non-transitory computer-readable storage medium storing one or more programs including instructions for performing the method of  claim 1 . 
     
     
         10 . A device for converting a dental image, the device comprising:
 a processor configured to:   receive the dental image;   receive a user input corresponding to a conversion area and a target type;   mask the conversion area of the dental image;   input the masked image to a pre-trained generative model; and,   obtaining an output dental image including a virtual image of at least one virtual tooth corresponding to the target type for the masked conversion area from the pre-trained generative model,   wherein, based on the target type including information for a treatment period type, the output dental image includes the virtual image of an intermediate treatment state of the at least one virtual tooth predicted based on an elapsed treatment period corresponding to the treatment period type,   wherein a number of the at least one virtual tooth is determined based on a size of the conversion area and a threshold size, and   wherein the threshold size is determined based on a position of the conversion area in the dental image.   
     
     
         11 . The device of  claim 10 , wherein the generated output dental image is generated via a generative sub-model corresponding to the target type of the pre-trained generative model. 
     
     
         12 . The device of  claim 10 , wherein the target type includes at least one of a first type corresponding to a normal state where no additional treatment is performed, a second type corresponding to a state recovered via the treatment, a third type corresponding to an abnormal state, and a fourth type corresponding to an under-treatment state. 
     
     
         13 . The device of  claim 10 , wherein the generative model includes a generative sub-model corresponding to the target type and is pre-trained based on a training image converted from an original dental image,
 wherein the training image for a generative sub-model corresponding to one target type is generated by converting the original dental image using type annotation information corresponding to a semantic area contained in the original dental image.   
     
     
         14 . The device of  claim 13 , wherein the semantic area corresponds to an individual tooth area contained in the dental image,
 wherein the type annotation information includes annotation information of a normal state type, annotation information of a post-treatment state type, annotation information of an under-treatment state type, or annotation information of an abnormal state type.   
     
     
         15 . The device of  claim 13 , wherein the training image for the generative sub-model corresponding to the target type determines an area-to-be-masked among the semantic areas based on the type annotation information,
 wherein the training image for the generative sub-model corresponding to the target type is generated from the original dental image by masking the area-to-be-masked.